Add lexer

Signed-off-by: trivernis <trivernis@protonmail.com>

Cargo.toml

@@ -14,7 +14,6 @@ repository = "https://github.com/Trivernis/charred-rs"
 [dependencies]
 thiserror = "1.0.24"
 log = "0.4.14"
-async-trait = "0.1.50"
 
 [dependencies.tokio]
 version = "1.5.0"

src/input_reader.rs

@@ -78,4 +78,9 @@ impl InputReader {
 
         Ok(())
     }
+
+    /// Returns the current index
+    pub fn index(&self) -> usize {
+        self.index
+    }
 }

src/lexer.rs

@@ -0,0 +1,40 @@
+use crate::error::TapeResult;
+use crate::input_reader::InputReader;
+use crate::token::{Token, TokenCheckerFn, UnknownToken};
+
+pub struct Lexer {
+    reader: InputReader,
+    checkers: Vec<TokenCheckerFn>,
+}
+
+impl Lexer {
+    /// Creates a new lexer with provided checker functions
+    pub fn new(reader: InputReader, checkers: Vec<TokenCheckerFn>) -> Self {
+        Self { reader, checkers }
+    }
+
+    /// Scans for tokens
+    pub async fn scan(&mut self) -> TapeResult<Vec<Token>> {
+        let mut tokens = Vec::new();
+
+        while !self.reader.check_eof().await {
+            let index = self.reader.index();
+            let mut found = false;
+
+            for checker_fn in &self.checkers {
+                if let Some(token) = checker_fn.as_ref()(&mut self.reader).await? {
+                    tokens.push(token);
+                    found = true;
+                    break;
+                } else {
+                    self.reader.seek_to(index).await?;
+                }
+            }
+            if !found {
+                tokens.push(Token::new(UnknownToken(self.reader.consume().await?)))
+            }
+        }
+
+        Ok(tokens)
+    }
+}

src/lib.rs

@@ -1,5 +1,6 @@
 pub mod error;
 pub mod input_reader;
+pub mod lexer;
 pub mod token;
 
 #[cfg(test)]

src/tests/mod.rs

@@ -1,2 +1,3 @@
 mod test_input;
+mod test_lexer;
 mod test_token;

src/tests/test_lexer.rs

@@ -0,0 +1,91 @@
+use crate::error::TapeResult;
+use crate::input_reader::InputReader;
+use crate::lexer::Lexer;
+use crate::token::{Token, TokenCheckerFn, UnknownToken};
+use std::io::Cursor;
+use std::sync::Arc;
+
+struct NumberToken(i32);
+struct StringToken(String);
+struct WhiteSpaceToken;
+
+async fn parse_number_token(reader: &mut InputReader) -> TapeResult<Option<Token>> {
+    let mut num = String::new();
+    while !reader.check_eof().await && reader.peek().await?.is_numeric() {
+        num.push(reader.consume().await?);
+    }
+    if num.is_empty() {
+        Ok(None)
+    } else {
+        Ok(Some(Token::new(NumberToken(num.parse::<i32>().unwrap()))))
+    }
+}
+
+async fn parse_whitespace_token(reader: &mut InputReader) -> TapeResult<Option<Token>> {
+    let mut count = 0;
+    while !reader.check_eof().await && reader.peek().await?.is_whitespace() {
+        reader.consume().await?;
+        count += 1;
+    }
+    if count > 0 {
+        Ok(Some(Token::new(WhiteSpaceToken)))
+    } else {
+        Ok(None)
+    }
+}
+
+async fn parse_string_token(reader: &mut InputReader) -> TapeResult<Option<Token>> {
+    let mut value = String::new();
+    while !reader.check_eof().await
+        && !reader.peek().await?.is_numeric()
+        && !reader.peek().await?.is_whitespace()
+    {
+        value.push(reader.consume().await?);
+    }
+    if value.is_empty() {
+        Ok(None)
+    } else {
+        Ok(Some(Token::new(StringToken(value))))
+    }
+}
+
+#[tokio::test]
+async fn it_scans() {
+    let checkers: Vec<TokenCheckerFn> = vec![
+        Arc::new(|reader| Box::pin(parse_number_token(reader))),
+        Arc::new(|reader| Box::pin(parse_whitespace_token(reader))),
+        Arc::new(|reader| Box::pin(parse_string_token(reader))),
+    ];
+    let input_reader = InputReader::new(Cursor::new("The Alphabet 12 ok"));
+    let mut lexer = Lexer::new(input_reader, checkers);
+    let tokens = lexer.scan().await.unwrap();
+    assert!(!tokens.is_empty());
+    let mut tokens = tokens.into_iter();
+    assert!(tokens.next().unwrap().is::<StringToken>());
+    assert!(tokens.next().unwrap().is::<WhiteSpaceToken>());
+    assert!(tokens.next().unwrap().is::<StringToken>());
+    assert!(tokens.next().unwrap().is::<WhiteSpaceToken>());
+    assert!(tokens.next().unwrap().is::<NumberToken>());
+    assert!(tokens.next().unwrap().is::<WhiteSpaceToken>());
+    assert!(tokens.next().unwrap().is::<StringToken>());
+}
+
+#[tokio::test]
+async fn it_falls_back_to_unknown() {
+    let checkers: Vec<TokenCheckerFn> = vec![
+        Arc::new(|reader| Box::pin(parse_number_token(reader))),
+        Arc::new(|reader| Box::pin(parse_string_token(reader))),
+    ];
+    let input_reader = InputReader::new(Cursor::new("The Alphabet 12 ok"));
+    let mut lexer = Lexer::new(input_reader, checkers);
+    let tokens = lexer.scan().await.unwrap();
+    assert!(!tokens.is_empty());
+    let mut tokens = tokens.into_iter();
+    assert!(tokens.next().unwrap().is::<StringToken>());
+    assert!(tokens.next().unwrap().is::<UnknownToken>());
+    assert!(tokens.next().unwrap().is::<StringToken>());
+    assert!(tokens.next().unwrap().is::<UnknownToken>());
+    assert!(tokens.next().unwrap().is::<NumberToken>());
+    assert!(tokens.next().unwrap().is::<UnknownToken>());
+    assert!(tokens.next().unwrap().is::<StringToken>());
+}

src/tests/test_token.rs

@@ -1,15 +1,12 @@
 use crate::error::TapeResult;
 use crate::input_reader::InputReader;
-use crate::token::{ProtoToken, Token};
-use async_trait::async_trait;
+use crate::token::Token;
 use std::io::Cursor;
 
 #[derive(Debug)]
 struct TestToken(i32);
 
-#[async_trait]
-impl ProtoToken for TestToken {
-    async fn try_parse(reader: &mut InputReader) -> TapeResult<Option<Token>> {
+async fn parse_test_token(reader: &mut InputReader) -> TapeResult<Option<Token>> {
     let mut num = String::new();
     while !reader.check_eof().await && reader.peek().await?.is_numeric() {
         num.push(reader.consume().await?);
@@ -19,23 +16,22 @@ impl ProtoToken for TestToken {
     } else {
         Ok(Some(Token::new(TestToken(num.parse::<i32>().unwrap()))))
     }
-    }
 }
 
 #[tokio::test]
 async fn it_parses() {
     let mut reader = InputReader::new(Cursor::new("128"));
-    let token = TestToken::try_parse(&mut reader).await.unwrap();
+    let token = parse_test_token(&mut reader).await.unwrap();
     assert!(token.is_some());
     let token = token.unwrap().try_into::<TestToken>().unwrap();
     assert_eq!(token.0, 128);
 
     let mut reader = InputReader::new(Cursor::new("string a12 24\n"));
-    let token = TestToken::try_parse(&mut reader).await.unwrap();
+    let token = parse_test_token(&mut reader).await.unwrap();
     assert!(token.is_none());
     reader.seek_to(8).await.unwrap();
 
-    let token = TestToken::try_parse(&mut reader).await.unwrap();
+    let token = parse_test_token(&mut reader).await.unwrap();
     assert!(token.is_some());
     let token = token.unwrap().try_into::<TestToken>().unwrap();
     assert_eq!(token.0, 12);

src/token.rs

@@ -1,13 +1,17 @@
 use crate::error::TapeResult;
 use crate::input_reader::InputReader;
-use async_trait::async_trait;
 use std::any::{Any, TypeId};
+use std::future::Future;
+use std::pin::Pin;
+use std::sync::Arc;
 
-#[async_trait]
-pub trait ProtoToken {
-    /// Tries parsing the token
-    async fn try_parse(reader: &mut InputReader) -> TapeResult<Option<Token>>;
-}
+pub type TokenCheckerFn = Arc<
+    dyn for<'a> Fn(
+            &'a mut InputReader,
+        ) -> Pin<Box<dyn Future<Output = TapeResult<Option<Token>>> + Send + 'a>>
+        + Send
+        + Sync,
+>;
 
 pub struct Token {
     inner: Box<dyn Any>,
@@ -38,3 +42,6 @@ impl Token {
         self.inner.as_ref().type_id() == TypeId::of::<T>()
     }
 }
+
+/// Parsed when no other matching token was found for the character
+pub struct UnknownToken(pub char);